Mar 19 2013

Evolved to Gamble

It is tempting to make arguments about how and why particular aspects of human psychology evolved. I will not be getting into an evolutionary psychology discussion in this post, I will just say I find such arguments offer a plausible framework for understanding human psychology, regardless of whether you think they can be scientifically tested.

A recent example is a paper by researchers at McMaster University. The title of their press release  reads: It’s in the cards: Human evolution influences gamblers’ decisions, study shows. 

The study actually says nothing about human evolution. It simply demonstrates an aspect of human behavior – the evolutionary explanation is pure speculation, not tested or demonstrated in the study itself.

Nevertheless, the study is interesting. It involved gambling behavior. Subjects were given money to gamble on which target would be next illuminated. The targets were illuminated at random, so any target had an equal chance of being the winning bet on any round.

The study found that subjects were more likely to change the target they bet on when their current target won, or after the target of another player won. In other words, they made their bets with the implicit assumption that the same target would not be illuminated twice in a row, even though that reasoning had absolutely no basis in reality. The subjects were not given any information that would make them think this was the case, in fact what they were told directly contradicts it. Also, the experience of playing would not have led to this conclusion, so again, it was something that was simply assumed by the subjects.

This is simply the well-known gambler’s fallacy – the sense that if you toss a fair coin and it comes up heads four or five times in a row, it is then “due” to come up tails. Of course, if the coin is truly fair then the next toss is still 50% likely to come up heads and 50% tails. Past events have no influence on the outcome of the next toss. Our intuitive sense is profoundly wrong in this case.

Gamblers sometime follow the opposite logic. On the roulette table, for example, they may play numbers that have come up before on the logic that they are on a streak, or that something is making them more likely to come up. At the same time they can play numbers that have not come up recently because they are “due” – and these contradictory bits of logic can coexist in the eager gambler.

The lead author, Jim Lyons, is quoted as saying:

“The results of our work suggest, perhaps for the first time, that certain aspects of problem gambling behaviour may be related to hard-wired, basic neurobiological factors related to how we direct our attention.”

This is a reasonable conclusion to draw from the study, and is consistent with prior research – these gambling behaviors appear to be hard-wired. Gambling itself appears to be culturally universal, but there isn’t much data on cultural vs innate factors in problem gambling across cultures. There is some evidence that the Chinese have greater problem gambling, with one review finding:

“Chinese gamblers exhibit severer cognitive errors, stronger risk-taking tendency and higher level of stress and depression, which explained the common findings of literature that Chinese have higher gambling and problem gambling rates than other ethnic groups.”

It seems that there are both universal and culture specific beliefs that contribute to gambling – those that involve luck and the illusion of control. In order to conclude that the gambler’s fallacy is evolved rather than learned or cultural it would need to be demonstrated to be culturally universal, at least to a degree. It seems the evidence is pointing in this direction, but is not completely confirmed at this point.

Where the discussion of this study gets into speculation – perfectly plausible and reasonable speculation, but speculation all the same, is when the press release attempts to explain why humans evolved the gambler’s fallacy heuristic. The press release says:

“They found that, like our ancestors, the gamblers relied on their past experiences to predict what might happen in the future. “

and then quoting psychologist Dan Weeks:

“Humans make rational decisions on a day-to-day basis based on experience. Think about someone picking apples in an orchard. Once the apples from the first tree are picked, it is a rational decision to move on to the next tree.”

It is certainly plausible that the foraging and hunting behavior of human ancestors contributed to the evolution of the gambler’s fallacy – don’t expect to get lucky in the same place twice, at least not soon. We just need to recognize that this is speculation, and not something we can derive from the evidence presented in this study.

In trying to understand the foibles of human cognition and behavior it is a useful idea that humans evolved in an environment that is very different from our modern civilization. We did not evolve with the stimulations and temptations presented by casinos. Similarly we did not evolve with virtually unlimited easy access to high caloric foods.

Psychological experiments have demonstrated a number of errors in thinking that people generally make. We have hyperactive agency detection, we have a “supersense” – a notion that living things have an essence, and even inanimate objects can possess an unseen property or significance, we tend to make assumptions about cause and effect, we seen patterns where they don’t exist, and find it difficult to accept that coincidences happen for no particular reason.

The gambler’s fallacy fits snugly into this suite of cognitive flaws and biases – the powerful assumption that past events influence future events, even when there is no plausible way that they can, and even when the system is designed to be as certain as possible that they don’t (which is the case with casinos and lotteries). We invent magical explanations, like luck, to explain what we feel to be right but know to be wrong.

It makes sense that this all evolved under some set of selective pressures. It would be interesting to study that directly, and perhaps someone is clever enough to figure out how. This current study, however, does not address the question.

35 responses so far

35 thoughts on “Evolved to Gamble”

  1. PharmD28 says:

    im in the process of finishing “thinking fast and slow” by daniel kahneman and the discussion about how we weigh probabilities different than what they really are is fascinating.

    I have only gambled truly 2 times, both at conferences in Vegas, both times played craps – both times I won about 250 dollars. For the most part I played the pass line bet (the best odds mostly as I understand)…..but I found myself taking riskier bets when I got going and had more money…

    I think both times I was quite lucky. The urge to win larger amounts in a shorter period of time is truly luring really…but not so much that I was willing personally to make bets that would give up all the money I made….guess I am a pretty cheap gambler 🙂

  2. cliff0079 says:

    I agree with the gamblers fallacy, but sometimes I think it is interpreted or used in somewhat of a belittling way to [some] gamblers. If we have a normal coin, it has a 50-50 chance for heads or tails (independent of past throws). If we flip a coin ten times it could certainly come up heads ten times in a row (1/1,024), and the probability for a heads on the next throw is still 50%, however. If we flipped it a billion times it should come to around 50-50 heads verse tails–given a normal coin. Now, I think some gamblers view that if a coin has flipped 50 heads in a row, the 51 throw may be “due” for a tails if in the long run it is going to adhere to the 50-50 probability (not that the probability for tails has change).

    This is unlike drawing cards from a deck of cards where the probabilities “actually do change” with each bet. As a result, it is logical to change bets given the discarded cards (e.g, your playing black jack with 52 cards and 75% of the face cards have been used, as a consequence the likelihood for the dealer winning have increased. So, you should reduce your bets–or leave the table).

  3. cliff – but it is still not due, because subsequent flips are not affected by the fact that the previous 50 flips were heads. It’s not as if they have to make up for that statistical anomaly with an anomaly in the opposite direction. They don’t. But it sure feels as if they do – that’s the fallacy.

    In fact, if a coin flipped heads 50 times in a row it would be more reasonable to suspect that it’s not a fair coin or fair flip, and that heads is more likely to come up if anything.

  4. Aidan says:

    I always loved the saying that the lottery is a tax on those who are bad at math. Though, I guess it is rather egotistical and this study demonstrate that the saying isn’t entirely correct. Someone could be good at math but simply be unable to apply it to the real world due to the cognitive biases that lead to the belief that past events influence future events. I guess a better saying would be the lotto is a tax on those who can’t overcome cognitive biases.

  5. tmac57 says:

    One thing that I would like to mention about the word ‘luck’,is that all of my life I have understood that word to refer to random good fortune,not actually possessing a quality of ‘luck’. It took me a long time to realize that some people actually see luck as a sort of mystical property that some people have ,and others don’t.
    I can’t tell you how many times I have heard the tale of the person who “always wins” when they go to the races,Vegas,etc. .When you point out that “sure it’s within the statistical range for that to happen”,especially in the short term,but that the long term odds are bound to catch up to virtually everyone if they keep on gambling where the built in odds are in favor of the house,they think that their case is exempt from probability.

  6. PharmD28 says:

    tma57….let me be clear…I am quite sure that if I kept playing, I would eventually lose all of those winnings!

  7. tmac57 says:

    PharmD28- Yes I got that,and I know your rep as a good skeptic,it just prompted me to write about a pet peeve of mine concerning the gambling mindset. No criticism of you intended at all.

  8. Murmur says:

    Jut reading what Steven is saying about seeing patterns where there aren’t and the belief in an “essence” got me thinking on where this Gamblers Fallacy could creep into other areas of our lives… in one place where people always talk about how lucky or unlucky they are, in their love lives. If we were to collate all the data for the number of people in this world who just happened to meet the right person for them in the very small sample of people they meet throughout their lives you would end up with a very small and improbable number.

    We then assume the world is set up to put those two people together, but really, the truth is there are hundreds, thousands, even tens of thousands of people out there who could be your perfect match, and even more who are close enough for people to settle with and perhaps break up with at some point.

  9. cliff0079 says:

    Novella- what I mean by due is that a fair coin in the long run should not show bias, otherwise it is not a fair coin. To your skepticism. 50 Heads in a row or 50 Tails in a row, or any combination in between all have the exact same probability. It’s always .5^n (for fair coins).

  10. Cliff – I understand, but that 0.5 probability will hold out no matter what point you start counting from, even if the prior 50 flips were all heads. Those prior flips will not be balanced out in the future making it more likely for tails to come up.

    In other words – if a true statistical outlier, like 50 heads, occurred, that is an unlikely event that would affect all statistics that include that data. It would not have to eventually balance out. In fact, it would require an equally unlikely event in the opposite direction which would take a very long time to occur, and may never occur over the length of observation.

    This is exactly why research often has to look at fresh data, and not include prior results, as you don’t want to carry forward a statistical outlier.

  11. SARA says:

    I think gambling is boring. The first time I played slots I won $75 on my second round and then walked away from casino to read in the bar. The second time I lost $20 and then went to the bar to read.

    On the other hand, I recognize that I also intuitively expect a tossed coin to change when it comes up heads 5 times.

    Obviously I’m not alone in just finding no lure in gambling. So, it seems to me there is some other component involved in gambling than the fallacy of odds.

  12. cliff0079 says:

    Novella–Yes, but how would you know the coin is still fair? You stated prior that you suspect a coin is not fair if it came up heads 50 times. A fair and not fair coin can do this. As a result, trials are needed to show the theoretical prediction.

  13. PharmD28 says:


    This is perhaps my favorite video ever by Tim Minchin…my wife thinks it is hilarious…I love her!

  14. PharmD28 says:

    I am not an expert in gambling…but I suspect many that feel they are particularly good or “lucky” at gambling suffer the same cherry picking of their own data as doctors used to during the days of blood letting….they give disproportional weight to when they win and less to when they lose…..

    oh, and to back track and quote that video:

    “well, I think your special, but you fall within a bell curve”….still laugh every time.

  15. cliff – by getting a fresh data set, as I said. If it was a statistical anomaly, new data should not replicate it. If it is an unfair coin, it may.

    This is how science works in general. If a study shows an effect, the effect may be real or may be a statistical fluke. A replication with fresh data is a good way to resolve the question.

    Sometimes researchers have “cheated” by doing a replication but including the original data. This has the effect of carrying forward any potential statistical flukes, and is not a true independent data set. Astrologers are famous for this.

  16. cliff0079 says:

    Okay, I understand your point better now and, I will concede to the majority of your points. (I think I am having trouble explaining a certain nuance, but that’s enough for today). Have a good day.

  17. pseudonymoniae says:

    Dr. Novella,

    Even accepting the premise that we can discuss evolutionary influences on psychology regardless of their testability/falsifiability, there are still issues with this kind of thinking. One key issue is the explicit assumption of evo psychological thought that specific features of human cognition and behavior represent adaptive features of our evolutionary past.

    The authors of this study note in their abstract:

    “We suggest that early humans developed specialized attentional systems to deal with non-random environmental contingencies…[which] are sometimes maladaptive in artificial environments in which the same contingencies do not hold”.

    In other words: there is something about gambling behavior that was adaptive in the past, but which is maladaptive now.

    But there are more plausible alternative explanations. For example, if susceptibility to probability errors or fallacies is not something which was selected against at any time in our evolutionary past, then such errors would operate under random genetic drift. In fact, there are lots of human faculties which are deficient for today’s world, but no reason why these abilities would have been required in the past. How is the “gambler’s fallacy” any different?

    These kinds of evo psych proposals also don’t attempt to differentiate between abilities on very specific tasks (e.g. foraging, hunting) and general purpose abilities like physical dexterity and fitness. If evolution acts on the latter, then we end up offering a flawed evolutionary explanation that relies upon a related, but distinct aspect of reality.

    It’s important to note that this study makes absolutely no attempt to relate its results to either “early humans” or evolutionary selection pressures, so it really doesn’t inform us about which option is more likely.

    The authors are speculating. The reporters are speculating. And I’m left wondering how it’s okay that we keep promoting so many flawed evo psych just-so stories that promulgate, rather than clarify, misconceptions about evolution.

  18. pseudo – I agree, that is worth a mention. There is an assumption of adaptation which is not warranted.

    I also agree, and tried to express in the article, that this study is not an evo psych study, says absolutely nothing about evo psych. It is simply by psychologists who add a layer of speculation about adaption to their discussion of the psychological effect they are documenting.

    I agree in this form it can be counterproductive. This is pseudo evo psych, if you will, and it should be criticized, but also I would not dismiss evo psych based upon abuse by those not even in the field.

  19. fitzylvania says:

    I have been reading the book Thinking, Fast and Slow, and they cover the same overall idea of human nature and trying to assign some greater power to random occurrences. I also really liked how they brought it back to basketball and how there is truly no such thing as a hot streak in terms of shooting, its just a long string of heads in a row eventually there would be a regression toward the mean with any extreme event or events. I have also had the thought in my head of the idea of assigning luck or good karma to everything that we do, it almost seems as a way of idealizing any good behavior that we do.

  20. sonic says:

    It seems the ‘gamblers fallacy’ might be in part due to our use of Bayesian analysis.

    I have seen coins flipped many times. They don’t do 27 in a row. My priors… That’s the Bayesian in me speaking there–

    Perhaps this should be added to the case for frequentist analysis.

  21. ConspicuousCarl says:


    Just for giggles and that other stuff…

    1/(2^50) =
    1/ 1,125,899,906,842,624

    US quarters in circulation, according to the first website I found with any number offered:

    So if even ONE trick coin has found its way into circulation, you are about a million times more likely, just by random, to end up holding that trick coin than you are to get 50 legitimate heads in a row.

    And then modify that with whatever your estimate might be of getting a trick coin when some jerk hands it to you specifically for the purpose of this experiment.

  22. lyonsjl says:

    Dr Novella

    Thank you for bring attention to our article through your blog. We appreciate your thoughtful comments as well as the interesting discussion your post has generated.

    Just to clarify a few points if I may. The evolutionary adaptation account for our results is related more to explaining the Inhibition of Return phenomenon to which we relate our observations of gambling behaviours. This idea, that that latencies in disengaging attention previously directed to a specific spatial location is related to foraging behaviours, was first posited by others several years ago (e.g., Klein and MacInnes, 1999, Psychological Science). It is certainly not the only explanation for the IOR effect but it is a compelling one. Our contribution, I think, was to suggest that the attention driven mechanisms that are thought to give rise to IOR may also be related to the expression of behaviours (both in action and belief) that are consistent with the Gambler’s fallacy. After all, when you think about it, that is exactly what the Gambler’s Fallacy is…an unwillingness to return to a location or course of action that has previously been explored and rewarded (or conversely, to go anew to an area that has not recently been rewarded).

    We do take great pains in the article to point out that this is not a generalized effect. Rather, we show that individuals who demonstrated the greatest susceptibility to the Gambler’s Fallacy also demonstrated the most pronounced IOR latency effects or, in other words, that IOR scores were a reliable predictor of betting behaviour. As to whether this is ultimately the result of an adaptive (or in the case of gambling, maladaptive) evolutionary progression is most certainly speculation on our part. We hope that others with more experience than us in the area of evolutionary psychology continue to speculate and further test the hypothesis.

    Thanks again for opening the discussion

  23. BillyJoe7 says:

    One thing to keep in mind is that the odds are valid only BEFORE the coin toss.
    The odds of throwing 10 heads in a row is 0.00098. With each toss that turns up yet another heads, the odds of throwing 10 heads in a row decreases. So that the odds of throwing 10 heads in row after the 9th throw also comes up heads is 0.5. When the 10th toss also comes up heads, the odds is 1. It has already happened.

  24. ConspicuousCarl says:

    One of the drivers of wacky modern gambling behavior might be the fact that we didn’t evolve to handle such enormous possible prizes, so the hyper motivation to try to win might be encouraging us to pile on the crazy in a desperate attempt to come up with an “answer”. A caveman was not likely to have the chance of winning a tree with a million apples on it. Gambling fallacies could just be stretched versions of things we do every day in ways that don’t stand out.

    We might find more similarly-strong motivation by considering natural death-avoidance scenarios, but then we are looking at loss aversion rather than gains, which are already shown to cause different behaviors.

    And speaking of apple-picking, I don’t find that to be too convincing as an explanation for “moving on” from a prior winning number. It could be related, but people also deny the probability of round or consecutive numbers in a single incident (e.g., they will say that 888 is less likely to occur than 672). It might be a tendency to assume that aesthetically elegant things are “special” and therefore unlikely.

  25. Kent Northcote says:

    Good Stuff Dr. Novella

    I’d just like to add that evolution of an appreciation for randomness wouldn’t, compared to the alternatives, undergo positive selection. Specifically, organisms defaulting to assume events have meaning and are linked should survive more often than organisms with a more laizzez faire attitude. Ignoring the breaking twigs outside is dangerous if it’s a predator whereas the energy spent checking on things just to be sure may produce anxiety but is less likely to render you into lunch. You touched upon this in your reference to our hyperactive agency detection device (HADD).

    The fact that gambling and the HADD are human universals points out that at the very least they arise from pre cultural forces.

  26. BillyJoe7 says:

    I used to think that a good point to make to someone who thinks they can win the lotto is to tell them to choose consecutive numbers (for example, when picking 6 numbers out of 40, choose the numbers 1,2,3,4,5,6). They will refuse to do so. You then explain that those 6 numbers have the same chance of coming up as the 6 numbers they have actually chosen (for example 2,9,12,17,34,37). The problem is that you cannot convince them that this is true.

  27. sonic says:

    I’m not sure playing the lottery is irrational in all cases.
    Say a person puts up 1 to win 1,000,000. His chance is 1 in 3,000,000 of winning, so it appears a bad bet. That’s because we analyze the utility of the money in a linear fashion.
    But the player sees it differently– he is comparing something he doesn’t care about ($1 won’t even buy a beer) to a lifetime of financial independence– something that is his wildest dream and unachievable any other way given his situation in life.

    We analyze the odds in a linear manner. But the rewards are not linear– they have an exponential aspect to them–

    It depends on how a person sees his financial future–

    If I think– ‘heck, I’ll be fine- I just invest this way and that… I can always earn large sums of money if I need to…’
    the lottery makes no sense.

    But if I think ‘my chances of making it are zero. I’ll continue to eat dog food for the rest of my life. This is certain.’
    You tell me I can up the odds to 1 in three million? Where do I sign up?

    Is that irrational?

  28. ConspicuousCarl says:


    That’s a good point and a reason why one of the common loss-aversion scenarios is not perfect. For some people, being $200 overdrawn is not really better than being $400 overdrawn. Either way, it’s the same problem.

    But any more than a dollar for a single lottery ticket and it might be different If you have ever been staring at a bank account with $2 left just before payday, you’d be glad you didn’t buy 3 lottery tickets that week. Dog food or not.

  29. SARA says:

    I’m not sure it’s entirely a non-linear choice.

    For example, I play the lottery maybe once a year. When it’s super high and I’m already somewhere that sells the tickets.

    Now, me winning the standard 5 million vs me winning 50 million are about as likely. And the impact in my life in both cases would be immediately positive with the difference amounting to yachts vs cruises.

    So why do I place so much more relative value on 50 million? As though 5 million just isn’t worth the $1.

    I must be thinking fairly linearly, since 50 is significantly more than 5.

  30. cliff says:


    Ha, funny. If only humans had a longer evolutionary history with math, especially statistics.

  31. sonic says:

    Conspicuous Carl-
    I’m thinking the costs aren’t really linear either– from what you say-
    It is a lot different to spend the dollar that won’t buy a beer than it is to spend the dollar that will complete the rent payment.
    In the one case you might not notice- in the other you might get booted from your abode.
    Not linear.
    The guy doesn’t come up with the $5, so when I get home all the furniture is gone.
    The people I’m living with decide it’s better to have 151 rum and weed than it is to have furniture.
    Perhaps the most rational decision I’ve ever witnessed. 🙂

    $200 is a smaller problem than $400. That problem isn’t linear either, as the biggest part of the problem is how to get $1. After that it’s just rinse and repeat- and by the time you’ve got $200, you know how to get $400.

    But I’m rambling about something I think we agree about.

    I believe if you calculate the expected pay-off before you wager, you are involved in what is called ‘investing’. 🙂

  32. BillyJoe7 says:

    I see the utility argument as trying to have your cake and eat it too. You can talk about the odds and how efectively you cannot win, but you can then go right ahead and play lotto anyway because of the utility value. Doesn’t wash. You spend a dollar on lotto and you’ve effectively flushed it down the toilet and missed out on a dollar worth of sweets or whatever. Any pleasure you have in anticipation of watching to see if your numbers come up is a delusion paid at the expense of the reality that you effectively have no chance.

  33. tmac57 says:

    Many people whom play the lottery on a regular basis,use the same “lucky set of numbers” ( ‘their’ numbers). I always thought that this was a tiny bit foolish,because even though it shouldn’t matter from the statistical point of view,there is some small chance that those numbers will come up during a drawing where they forgot to play,thus causing them a great deal of psychological harm. And yes,I do realize that them forgetting to play those numbers is no different than them forgetting to play any random set of numbers,but I would guess that that would be nearly impossible for them to reconcile.
    Of course the danger of this happening is trivially small even if it is non-zero. It’s probably more worrisome for them to actually win the lottery,if we are to believe all of the tragic stories that you hear about winner’s misfortunes 🙂

  34. etatro says:

    If the author’s early training was in biology, he probably habitually frames everything within the theory of evolution. That is a starting point for me, usually, when I conceptualize things, then narrow it down to the specific question at hand, then broaden it back out. I almost always try to think about how natural selection or genetic drift may have led to the phenomena that I’m observing. Any good scientist studying living things (even behavior) would. Probably the author (Jim Lyons) got overzealous on the matter while talking with the PR manager … who then riffed and generated this article. Could be worse.

  35. Kent Northcote says:

    Evolved to Gamble?

    The phenomenon of hyperbolic discounting lends credence to the idea that gambling preys upon our evolutionary ingrained tendencies. Well portrayed by George Ainslie in ‘Breakdown of Will’, Dr. Ainslie, a psychiatrist and addiction researcher who works with the VA population, hyperbolic discounting is the phenomenon of the value of a good changing hyperbolically the closer one gets to the time of realizing that reward.
    For instance, an alcoholic may hold very little value for taking a drink ten days from now and may value not drinking in ten days much more. As the day approaches and particularly when the day is imminent, the value of the drink rockets up, overtaking the value of abstinence, and drinking ensues.
    As to how this phenomenon reinforces the idea that gambling addiction may have evolutionary roots we can turn to the idea that for most animals, lacking long term memory and symbolic representation as they do, the immediate awards far outweigh what may come in the future. My cat probably doesn’t even have a concept of next week, much less the rest of his life.
    Since our mental life, as multi-varied as it is with essentially an infinite number of conscious experiences available to us, thanks in large part to the number of states achievable with our thalamocortical system, is nevertheless underpinned by our affective machinery and since that affective machinery, from the brain stem, notably the periaqueductal grey, and the superior colliculus and on through to the limbic system, such as it is, is conserved across all mammals, and to a large extent, all vertebrates, we should not be too surprised to see the same types of emotionally based actions in ourselves as we do in ‘lesser’ animals. In short gambling addictions, and most addictions for that matter, are probably behavioral fossils.

    Kent Northcote MD

Leave a Reply